import the libraries¶
In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
### so that u dont have warnings
from warnings import filterwarnings
filterwarnings('ignore')
read the dataset and check the info of dataset¶
In [2]:
#read dataset
df=pd.read_csv('D:\python project\Zomato Data Analysis/zomato.csv')
df.head()
Out[2]:
In [4]:
df.shape
Out[4]:
In [5]:
df.info()
In [6]:
df.isnull().sum()
Out[6]:
cleaning 'rate' column of data¶
analyst 'rate'
In [8]:
#drop the missing value in 'rate'column
df.dropna(axis='index',subset=['rate'],inplace=True)# have to remove it in a horizontal way,and use subset parameter and replace data
In [10]:
df['rate'].unique()
Out[10]:
In [11]:
def split(x):#define function(parameter)
return x.split('/')[0]#use this function
In [12]:
df['rate']=df['rate'].apply(split)#use this function to excute each rows or each columns
In [13]:
df['rate'].unique()
Out[13]:
In [14]:
df.replace('NEW',0,inplace=True)#use 0 to replace new
df.replace('-',0,inplace=True)#use 0 to replace -
In [15]:
df['rate']=df['rate'].astype(float)#change a type of this colum
calculate avg rating of each resturant
In [17]:
df.groupby('name')['rate'].mean()
Out[17]:
In [19]:
df.groupby('name')['rate'].mean().to_frame()
Out[19]:
In [20]:
df_rate=df.groupby('name')['rate'].mean().to_frame().reset_index()
df_rate
Out[20]:
rename the columns
In [22]:
df_rate.columns=['restaurant','avg_rating']
df_rate.head(20)
Out[22]:
analyst 'rate' and 'book_table' and 'online_order'¶
rating distribution¶
In [24]:
sns.set_style(style='whitegrid')#白色网格形式
sns.distplot(df_rate['avg_rating'])
Out[24]:
resume:Almost more than 50 percent of restaurants has rating between 3 and 4. Restaurants having rating more than 4.5 are very rare
Which are the top restaurant chains in Bangaluru?¶
In [28]:
chains=df['name'].value_counts()[0:20]
chains
Out[28]:
In [30]:
plt.figure(figsize=(10,7))
chains=df['name'].value_counts()[0:20]
sns.barplot(x=chains,y=chains.index,palette='deep')#chains=number,chains.index=name of res
plt.title("Most famous restaurants chains in Bangaluru")
plt.xlabel("Number of outlets")
Out[30]:
How many of the restuarants do not accept online orders?¶
In [31]:
x=df['online_order'].value_counts()
x
Out[31]:
In [32]:
!pip install plotly
In [33]:
import plotly.express as px
In [34]:
x=df['online_order'].value_counts()
labels=['accepted','not accepted']
fig = px.pie(df, values=x, names=labels,title='Pie chart')
fig.show()
In [35]:
# another way to draw
x=df['online_order'].value_counts()
labels=['accepted','not accepted']
plt.pie(x,explode=[0.0,0.1],autopct='%1.1f%%')#Keep one decimal place
Out[35]:
What is the ratio b/w restaurants that provide and do not provide table booking ?¶
In [36]:
x=df['book_table'].value_counts()
x
Out[36]:
In [37]:
import plotly.graph_objs as go
from plotly.offline import iplot
In [38]:
x=df['book_table'].value_counts()
labels=['not book','book']
In [39]:
trace=go.Pie(labels=labels, values=x,
hoverinfo='label+percent', textinfo='value',
textfont=dict(size=25),
pull=[0, 0, 0,0.2, 0]
)
iplot([trace])
In [40]:
#another way to draw
x=df['book_table'].value_counts()
labels=['not book','book']
plt.pie(x,explode=[0.0,0.1],autopct='%1.1f%%')
Out[40]:
cleaning 'rest_type' column of data¶
In [43]:
df['rest_type'].isna().sum()
Out[43]:
In [44]:
df['rest_type'].dropna(inplace=True)
analyst res_type¶
In [45]:
len(df['rest_type'].unique())#have 87 restanrants types
Out[45]:
different type of restaurants¶
In [62]:
#data is represented in form of percentage
(df['rest_type'].value_counts()/len(df))*100
Out[62]:
what is the top 20 of restanrant types¶
In [46]:
plt.figure(figsize=(20,12))
df['rest_type'].value_counts().nlargest(20).plot.bar(color='red')
plt.gcf().autofmt_xdate()
In [47]:
# another way to draw
trace1 = go.Bar(
x = df['rest_type'].value_counts().nlargest(20).index,
y = df['rest_type'].value_counts().nlargest(20),
name= 'rest_type')
iplot([trace1])
highest voted restaurant¶
In [51]:
df.groupby('name')['votes'].sum().nlargest(10).plot.bar()
Out[51]:
In [49]:
#another way to draw
trace1 = go.Bar(
x = df.groupby('name')['votes'].sum().nlargest(10).index,
y = df.groupby('name')['votes'].sum().nlargest(10))
iplot([trace1])
analyst 'location'¶
familier with 'location'column¶
In [54]:
df.isnull().sum()
Out[54]:
In [55]:
df.groupby('location')['name'].unique()
Out[55]:
how namy restaurants in difference district¶
In [63]:
location=[] #key
restaurant=[] #value
for key,location_df in df.groupby('location'):
location.append(key)
restaurant.append(len(location_df['name'].unique()))
In [64]:
df_total=pd.DataFrame(zip(location,restaurant))
df_total.columns=['location','restaurant']# rename
df_total.head()
Out[64]:
In [65]:
df_total.set_index('location',inplace=True)
df_total.sort_values(by='restaurant').tail(10)# display the last 10 of dataset
Out[65]:
In [66]:
df_total.sort_values(by='restaurant').tail(10).plot.bar()
Out[66]:
In [67]:
df_total.sort_values(by='restaurant').tail(10).index
Out[67]:
In [68]:
trace1 = go.Bar(
x = df_total['restaurant'].nlargest(10).index,
y = df_total['restaurant'].nlargest(10),
name= 'Priority')
iplot([trace1])
analyst 'cuisines'¶
the top 10 of number of cuisines¶
In [69]:
cuisines=df['cuisines'].value_counts()[:10]#top 10 of number of cuisines
cuisines
Out[69]:
In [70]:
sns.barplot(cuisines,cuisines.index)
plt.xlabel('Count')
plt.title("Most popular cuisines of Bangalore")
Out[70]:
In [71]:
#another way to draw
cuisines=df['cuisines'].value_counts()[:10]
trace1 = go.Bar(
x = cuisines.index,
y = cuisines,
name= 'Cuisines')
iplot([trace1])
resume: We can observe that North Indian,chinese,South Indian and Biriyani are most common.
cleaning 'approx cost(for 2 people)'column¶
In [72]:
len(df['approx_cost(for two people)'].value_counts())
Out[72]:
In [73]:
df['approx_cost(for two people)'].isna().sum()
Out[73]:
In [77]:
df.dropna(axis=0,subset=['approx_cost(for two people)'],inplace=True)
# axis=0,是指删除行；subset指的是限定函数作用范围，只在[‘col1’,‘col2’…]列内
In [78]:
df['approx_cost(for two people)'].isna().sum()
Out[78]:
In [79]:
df['approx_cost(for two people)'].unique()
Out[79]:
In [80]:
df['approx_cost(for two people)']=df['approx_cost(for two people)'].apply(lambda x:x.replace(',',''))
#把上面数值中有逗号的去掉逗号
In [81]:
df['approx_cost(for two people)'].unique()
Out[81]:
In [82]:
df['approx_cost(for two people)']=df['approx_cost(for two people)'].astype(int)
In [83]:
df['approx_cost(for two people)'].dtype
Out[83]:
analyst 'approx_cost(for two people)'¶
distribution for 2 people¶
In [85]:
df['approx_cost(for two people)']
Out[85]:
In [84]:
sns.distplot(df['approx_cost(for two people)'])
Out[84]:
In [90]:
#another way to draw
px.histogram(df, x="approx_cost(for two people)")
resume:1.if customers have 2,they usully spend the cost about 500 2.most of the price lies between in a range of under 1000,it means most are affordable & very few are luxurious
cheapest rate for 2 people¶
In [86]:
df['approx_cost(for two people)'].min()
Out[86]:
highest rate for 2 people¶
In [87]:
df['approx_cost(for two people)'].max()
Out[87]:
what is the highest restarant infor¶
In [88]:
df[df['approx_cost(for two people)']==6000]
Out[88]:
In [89]:
df[df['approx_cost(for two people)']==6000]['name']
Out[89]:
cost vs rating¶
In [91]:
plt.figure(figsize=(10,7))
sns.scatterplot(x="rate",y='approx_cost(for two people)',hue='online_order',data=df)
plt.show()
resume:the higher rate ,the higher cost (for 2 people)
Is there any difference between price of restaurants accepting and not accepting online orders for 2 people?¶
In [95]:
fig = px.box(df,x='online_order',y='approx_cost(for two people)')
fig.show()
most expensive and most favirate dish with approx cost for 2 people¶
In [96]:
df[df['approx_cost(for two people)']==6000].loc[:,('name','cuisines','dish_liked')]
# or df[df['approx_cost(for two people)']==6000][['name','cuisines','dish_liked']]
Out[96]:
Top 10 Most Expensive restaurant with approx cost for 2 people¶
In [109]:
data=df.copy()
In [110]:
data.dtypes
Out[110]:
In [111]:
data.set_index('name',inplace=True)
In [112]:
data['approx_cost(for two people)'].nlargest(10).plot.bar()
#最贵的餐厅里有多少道菜最贵，价格是多少
Out[112]:
In [103]:
# another way to draw
trace1 = go.Bar(
x = data['approx_cost(for two people)'].nlargest(10).index,
y = data['approx_cost(for two people)'].nlargest(10),
name= 'Priority')
iplot([trace1])
Top 10 Cheapest restaurant with approx cost for 2 people¶
In [113]:
data['approx_cost(for two people)'].nsmallest(10).plot.bar()
Out[113]:
In [114]:
trace1 = go.Bar(
x = data['approx_cost(for two people)'].nsmallest(10).index,
y = data['approx_cost(for two people)'].nsmallest(10),
name= 'Priority')
iplot([trace1])
Top 10 Cheapest restaurant location wise with approx cost for 2 people¶
In [115]:
data.set_index('location',inplace=True)
data['approx_cost(for two people)'].nsmallest(10)
Out[115]:
all the restautant that are below than 500(budget hotel)¶
In [116]:
data[data['approx_cost(for two people)']<=500]
Out[116]:
how many restaurant ,the cost<=500 for two people¶
In [117]:
df_budget=data[data['approx_cost(for two people)']<=500].loc[:,('approx_cost(for two people)')]
df_budget=df_budget.reset_index()
df_budget.head()
Out[117]:
In [118]:
df_budget['approx_cost(for two people)'].value_counts().plot.bar()
Out[118]:
In [119]:
trace1 = go.Bar(
x = df_budget['approx_cost(for two people)'].value_counts().index,
y = df_budget['approx_cost(for two people)'].value_counts(),
name= 'Priority')
iplot([trace1])
resume:we can see the dishes of cost around 300-400 are maximum in count
Total no. of Restaurants that have better rating >4 and that are under budget too ie less than 500¶
In [120]:
df_new=df[(df['rate']>=4) & (df['approx_cost(for two people)']<=500)]
len(df_new['name'].unique())
Out[120]:
total affordable restarant at different location¶
In [121]:
location=[] #key
total=[]#value
for loc,location_df in df_new.groupby('location'):
location.append(loc)
total.append(len(location_df['name'].unique()))
In [122]:
len(location)
Out[122]:
In [123]:
len(total)
Out[123]:
In [124]:
location_df=pd.DataFrame(zip(location,total))
location_df.columns=['location','restaurant']
location_df.set_index('location',inplace=True)
location_df.head(20)
Out[124]:
In [125]:
location_df['restaurant'].nlargest(10).plot.bar()
plt.gcf().autofmt_xdate()
plt.ylabel('Total restaurants')
Out[125]:
In [126]:
#another way to draw
trace1 = go.Bar(
x = location_df['restaurant'].nlargest(10).index,
y = location_df['restaurant'].nlargest(10),
name= 'Priority')
iplot([trace1])
to visualise what are the names of restaurants
In [127]:
location=[]
total=[]
for loc,location_df in df_new.groupby('location'):
location.append(loc)
total.append(location_df['name'].unique())
In [128]:
afford=pd.DataFrame(zip(location,total))
afford.columns=['location','res_names']
afford.set_index('location',inplace=True)
afford.head()
Out[128]:
finding best budget restaurants in any location¶
In [129]:
def return_budget(location,restaurant):
budget=df[(df['approx_cost(for two people)']<=400) & (df['location']==location) &
(df['rate']>4) & (df['rest_type']==restaurant)]
return(budget['name'].unique())
In [130]:
return_budget('BTM',"Quick Bites")
Out[130]:
which are the foodie areas?¶
In [131]:
plt.figure(figsize=(10,7))
Restaurant_locations=df['location'].value_counts()[:20]
sns.barplot(Restaurant_locations,Restaurant_locations.index)
Out[131]:
In [132]:
#another way to draw
Restaurant_locations=df['location'].value_counts()[:20]
trace1 = go.Bar(
x = Restaurant_locations.index,
y = Restaurant_locations,
name= 'Priority')
iplot([trace1])
resume:We can see that BTM,HSR and Koranmangala 5th block has the most number of restaurants
geographical analysis¶
I need Latitudes & longitudes for each of the place for geaographical Data analysis,so to fetch lat,lon of each place,use Geopy
In [133]:
locations=pd.DataFrame({"Name":df['location'].unique()})
In [134]:
locations['new_Name']='Bangalore '+locations['Name']
In [135]:
locations.head()
Out[135]:
In [136]:
!pip install geopy
In [140]:
from geopy.geocoders import Nominatim
In [143]:
lat_lon=[]
geolocator=Nominatim(user_agent="app")
for location in locations['Name']:
location = geolocator.geocode(location)
if location is None:
lat_lon.append(np.nan)
else:
geo=(location.latitude,location.longitude)
lat_lon.append(geo)
locations['geo_loc']=lat_lon
In [144]:
locations.head()
Out[144]:
In [145]:
locations.to_csv('zomato_locations.csv',index=False)
We have found out latitude and longitude of each location listed in the dataset using geopy.
In [146]:
# This is used to plot maps
Rest_locations=pd.DataFrame(df['location'].value_counts().reset_index())
In [147]:
Rest_locations.columns=['Name','count']
Rest_locations.head()
Out[147]:
In [148]:
# now combine both the dataframes
locations.head()
Out[148]:
In [149]:
Restaurant_locations=Rest_locations.merge(locations,on='Name',how="left").dropna()
Restaurant_locations.head()
Out[149]:
In [150]:
Restaurant_locations.shape
Out[150]:
In [151]:
Restaurant_locations['count'].max()
Out[151]:
In [152]:
type(Restaurant_locations['geo_loc'][0])
Out[152]:
In [153]:
def generateBaseMap(default_location=[12.97, 77.59], default_zoom_start=12):
base_map = folium.Map(location=default_location, zoom_start=default_zoom_start)
return base_map
In [154]:
len(Restaurant_locations['geo_loc'])
Out[154]:
In [155]:
Restaurant_locations.isna().sum()
Out[155]:
In [156]:
Restaurant_locations['geo_loc'][0][0]
Out[156]:
In [157]:
Restaurant_locations['geo_loc'][0][1]
Out[157]:
In [158]:
np.array(Restaurant_locations['geo_loc'])
Out[158]:
In [159]:
lat,lon=zip(*np.array(Restaurant_locations['geo_loc']))
In [160]:
Restaurant_locations['lat']=lat
Restaurant_locations['lon']=lon
Restaurant_locations.head()
Out[160]:
In [161]:
!pip install folium
In [162]:
import folium
from folium.plugins import HeatMap
basemap=generateBaseMap()
In [163]:
basemap
Out[163]:
In [164]:
Restaurant_locations[['lat','lon','count']].values.tolist()
Out[164]:
In [165]:
HeatMap(Restaurant_locations[['lat','lon','count']].values.tolist(),zoom=20,radius=15).add_to(basemap)
Out[165]:
In [166]:
basemap
Out[166]:
resume:It is clear that restaurants tend to concentrate in central bangalore area.¶ The clutter of restaurants lowers are we move away from central. So,potential restaurant entrepreneurs can refer this and find out good locations for their venture
heatmap of north and south indian restaurants¶
In [167]:
df.head()
Out[167]:
In [168]:
df2= df[df['cuisines']=='North Indian']
df2.head()
Out[168]:
In [169]:
north_india=df2.groupby(['location'],as_index=False)['url'].agg('count')
north_india.columns=['Name','count']
north_india.head()
Out[169]:
In [170]:
north_india=north_india.merge(locations,on="Name",how='left').dropna()
In [171]:
north_india.head()
Out[171]:
In [172]:
north_india['lan'],north_india['lon']=zip(*north_india['geo_loc'].values)
In [173]:
north_india.drop(['geo_loc'],axis=1)
Out[173]:
In [174]:
basemap=generateBaseMap()
HeatMap(north_india[['lan','lon','count']].values.tolist(),zoom=20,radius=15).add_to(basemap)
basemap
Out[174]:
What about South Indian cuisines?
In [175]:
df3= df[df['cuisines']=='South Indian']
south_india=df2.groupby(['location'],as_index=False)['url'].agg('count')
south_india.columns=['Name','count']
south_india=south_india.merge(locations,on="Name",how='left').dropna()
south_india['lan'],south_india['lon']=zip(*south_india['geo_loc'].values)
In [176]:
south_india=south_india.drop(['geo_loc'],axis=1)
In [177]:
south_india.head()
Out[177]:
In [178]:
basemap=generateBaseMap()
HeatMap(south_india[['lan','lon','count']].values.tolist(),zoom=20,radius=15).add_to(basemap)
basemap
Out[178]:
Which are the most popular casual dining restaurant chains?¶
In [179]:
df_1=df.groupby(['rest_type','name']).agg('count')
datas=df_1.sort_values(['url'],ascending=False).groupby(['rest_type'],
as_index=False).apply(lambda x : x.sort_values(by="url",ascending=False).head(3))['url'].reset_index().rename(columns={'url':'count'})
In [180]:
datas
Out[180]:
In [181]:
df_1=df.groupby(['rest_type','name']).agg('count')
df_1
Out[181]:
In [182]:
df_1.sort_values(['url'],ascending=False)
Out[182]:
In [184]:
df_1.sort_values(['url'],ascending=False).groupby(['rest_type'],as_index=False).apply(lambda x : x.sort_values(by="url",ascending=False))
Out[184]:
In [185]:
df_1.sort_values(['url'],ascending=False).groupby(['rest_type'],as_index=False).apply(lambda x : x.sort_values(by="url",ascending=False))['url']
Out[185]:
In [186]:
df_1.sort_values(['url'],ascending=False).groupby(['rest_type'],as_index=False).apply(lambda x : x.sort_values(by="url",ascending=False))['url'].reset_index()
Out[186]:
In [187]:
dataset=df_1.sort_values(['url'],ascending=False).groupby(['rest_type'],
as_index=False).apply(lambda x : x.sort_values(by="url",ascending=False))['url'].reset_index().rename(columns={'url':'count'})
In [188]:
dataset
Out[188]:
In [189]:
casual=dataset[dataset['rest_type']=='Casual Dining']
casual
Out[189]:
We can see tht Empire restaurant,Beijing bites and Mani's dum biriyani are the most popular casual dining restaurant chains in Bangalore. We will inspect them further...
wordcloud of dishes liked cy cuisines¶
In [190]:
!pip install wordcloud
In [191]:
from wordcloud import WordCloud
In [192]:
df.head()
Out[192]:
In [193]:
df['update_dish_liked']=df['dish_liked'].apply(lambda x : x.split(',') if type(x)==str else [''])
In [194]:
df.head()
Out[194]:
In [195]:
df['rest_type'].value_counts()[:9].index
Out[195]:
create wordcloud for each and every restaurant
In [196]:
from wordcloud import WordCloud, STOPWORDS
In [197]:
df.isna().sum()
Out[197]:
In [198]:
df.dropna(axis='index',subset=['rest_type'],inplace=True)
df.dropna(axis='index',subset=['dish_liked'],inplace=True)#axis='index相当于axis=0'
In [199]:
df.isna().sum()
Out[199]:
wordcloud for 1 restaurant
In [200]:
data=df[df['rest_type']=='Quick Bites']
In [201]:
data['dish_liked']
Out[201]:
In [202]:
stopwords=set(STOPWORDS)# 设置停用词
In [203]:
dishes=''
plt.figure(figsize=(10,20))
for word in data['dish_liked']:
words=word.split()
# Converts each token into lowercase
for i in range(len(words)):
words[i] = words[i].lower()
dishes=dishes+ " ".join(words)+" "
wordcloud = WordCloud(max_font_size=None, background_color='white', collocations=False,stopwords = stopwords,width=1500, height=1500).generate(dishes)
plt.imshow(wordcloud)
plt.axis("off")
Out[203]:
analysing reviews of particular restaurant¶
In [221]:
df.head()
Out[221]:
In [222]:
df['reviews_list'][0]
Out[222]:
In [223]:
data=df['reviews_list'][0].lower()
data
Out[223]:
In [224]:
import re
data2=re.sub('[^a-zA-Z]', ' ',data)
data2
Out[224]:
In [225]:
data3=re.sub('rated', ' ',data2)#use space to replace rated
data3
Out[225]:
In [226]:
data4=re.sub('x',' ',data3)#use space to replace x
data4
Out[226]:
In [227]:
re.sub(' +',' ',data4)#use one space to replace multiple spaces
Out[227]:
analyse review of a particular restaurant
In [228]:
dataset=df[df['rest_type']=='Quick Bites']
In [230]:
type(dataset['reviews_list'][3])
Out[230]:
In [231]:
total_review=' '
for review in dataset['reviews_list']:
review=review.lower()
review=re.sub('[^a-zA-Z]', ' ',review)
review=re.sub('rated', ' ',review)
review=re.sub('x',' ',review)
review=re.sub(' +',' ',review)
total_review=total_review + str(review)
In [232]:
wordcloud = WordCloud(width = 800, height = 800,
background_color ='white',
stopwords = stopwords,
min_font_size = 10).generate(total_review)
# plot the WordCloud image
plt.figure(figsize = (8, 8))
plt.imshow(wordcloud)
plt.axis("off")
Out[232]:
In [233]:
def importance(restaurant):
dataset=df[df['rest_type']==restaurant]
total_review=' '
for review in dataset['reviews_list']:
review=review.lower()
review=re.sub('[^a-zA-Z]', ' ',review)
review=re.sub('rated', ' ',review)
review=re.sub('x',' ',review)
review=re.sub(' +',' ',review)
total_review=total_review + str(review)
wordcloud = WordCloud(width = 800, height = 800,
background_color ='white',
stopwords = set(STOPWORDS),
min_font_size = 10).generate(total_review)
# plot the WordCloud image
plt.figure(figsize = (8, 8))
plt.imshow(wordcloud)
plt.axis("off")
In [234]:
importance('Quick Bites')
In [ ]: